Preview only show first 10 pages with watermark. For full document please download

Data Storage Ii.

   EMBED


Share

Transcript

Data Storage II. Prof. Ing. Pavel Tvrdík CSc. Ing. Jiří Kašpar Department of Computer Systems Faculty of Information Technology Czech Technical University in Prague © Pavel Tvrdík, Jiří Kašpar, 2011 Advanced Computer System Architectures, MI-POA, 03/2011, Lecture 4 https://edux.fit.cvut.cz/courses/MI-POA Evropský sociální fond Praha & EU: Investujeme do vaší budoucnosti Pavel Tvrdík, Jiří Kašpar (ČVUT FIT) Data Storage II. MI-POA, 2011, Lecture 4 1/22 Volume Management Volume Management • Volume management is a mapping of Logical Volume blocks to a set of Physical Disks. It is implemented using RAID and disk partitioning techniques. We can find volume management in: • storage controllers, • operating systems as o standalone component, o integrated in file system (Tru64 AdvFS, Solaris ZFS, ...) o integrated in database (Oracle ASM) Pavel Tvrdík, Jiří Kašpar (ČVUT FIT) Data Storage II. MI-POA, 2011, Lecture 4 2/22 Volume Management Storage Adapters and Controllers Storage Adapters and Controllers • Adapter is a I/O bus to Storage bus transformer (no Volume Management functions) • Controller is an Adapter with integrated Volume Management functions Internal or External Storage Server I/O Bus Adapter/ Controller CPU Storage Bus Server I/O Bus CPU Storage Bus Adapter Storage Controller Storage backend Bus External Storage Pavel Tvrdík, Jiří Kašpar (ČVUT FIT) Data Storage II. MI-POA, 2011, Lecture 4 3/22 Volume Management Volume Management Terminology Volume Management Terminology • • • • Physical Disk (=Physical Volume) Logical Volume Group is a set of physical disks. Spare Set is a set of spare (physical) disks. Logical Volume is an independent disk address space presented to hosts. • Logical Unit Number (LUN) is a number used to identify Logical Volume to Host. • Host is a computer system. Pavel Tvrdík, Jiří Kašpar (ČVUT FIT) Data Storage II. MI-POA, 2011, Lecture 4 4/22 Volume Management Advanced Volume Management Functions Advanced Volume Management Functions Three methods for creating physical or logical copy of a Logical Volume: • Volume Clone starts operation with full consistent RAID 1 set, and ends with two independent Volumes. • Volume Snapshot starts with single copy of data and saves old copy of modified data to a Snapshot. Unchanged data remains in a single copy. • Volume SnapClone starts with single copy of data as a snapshot, also saves old copy of modified data to snapclone, but in the background performs a copy of all data to the Snapclone. Pavel Tvrdík, Jiří Kašpar (ČVUT FIT) Data Storage II. MI-POA, 2011, Lecture 4 5/22 Multipathing Multipathing Server & OS • Both paths can be active, behavior depends on both: Adapter Adapter o Controller and o Operating System. • Usually one path is active, the second passive. SAN SAN Adapter Adapter Storage Controller Pavel Tvrdík, Jiří Kašpar (ČVUT FIT) Data Storage II. MI-POA, 2011, Lecture 4 6/22 Disk Storage Summary Disk Storage Summary • Direct Attached Storage (DAS): disks (in a box or external) are connected directly to storage adapter/controller on server I/O bus. Volume management is optional. Access on block level. • Enterprise Storage Architecture: disks are connected at Storage backend storage bus, server(s) to the front-end storage bus. Volume Management is provided by storage. Access on block level. • Network Attached Storage (NAS) Architecture: Server communicates with storage using file access protocols (NFS, SMB) on top of TCPIP transport. NAS server runs embedded Linux/Windows OS with full storage stack. Access on filesystem level. Pavel Tvrdík, Jiří Kašpar (ČVUT FIT) Data Storage II. MI-POA, 2011, Lecture 4 7/22 Tape Technologies Tape Technologies • Tape Drive Principles • Tape Libraries • Virtual Tape Libraries Pavel Tvrdík, Jiří Kašpar (ČVUT FIT) Data Storage II. MI-POA, 2011, Lecture 4 8/22 Tape Technologies Tape Device Tape Device Tape Device is a sequential block oriented device with variable block size. A sequences of data blocks can be separated by special control blocks called “Tape Marks”. Start and end of physical tape is detected using mechanical or optical mark on a tape. Pavel Tvrdík, Jiří Kašpar (ČVUT FIT) Data Storage II. MI-POA, 2011, Lecture 4 9/22 Tape Technologies Tape Terminology Tape Terminology Write/Read Heads • The contact point at which the actual electronic data changes are written onto physical tape media. Servo Head • Read only and align the read/write heads by reading pre-recorded information tracks form the tape Tracks • Physical “lines” of data on the tape Format • The physical & logical structure (layout) of data on the media. Bit density • How many “bits” are recorded per inch (bpi) Streaming • Maintaining forward motion of the tape by supplying data fast enough Compression ratio • • How compressible a piece of data – affects the capacity of a piece of media. Quote 2:1 but 1.4:1 is more typical. Append • The ability to write multiple backup sessions onto a single tape thereby making full use of the capacity of the tape. Pavel Tvrdík, Jiří Kašpar (ČVUT FIT) Data Storage II. MI-POA, 2011, Lecture 4 10/22 Tape Technologies Tape Terminology Tape Terminology (continued) Cartridge memory • A 4K “flash” memory chip located in every piece of LTO media – assists fast search by keeping a tape directory, also used to track media quality. Fast search • A typical full tape backup take around 2 hours, yet we can recover a single file from tape in around 70 seconds – using “fast search” by placing the drive in a special search mode. WORM • Write once read many – the ability of the media/drive to prevent overwriting data on the tape – for use in “tamper proof” data recording/compliance Interface • • The physical connect which connects the tape drive to the “system” or library ( SCSI, Fibre Channel, USB, SAS) Backwards compatibility • The ability of one “generation” of tape drive to write/read the format of previous generations of tape drive (example later) Barcode label • A label which a barcode reader can use inside an autoloader or Library to uniquely identify a piece of media. Autoloader • A device with a robot for inserting media, but with only one tape drive Library • A device with one or more robots for inserting media and more than one tape drive Pavel Tvrdík, Jiří Kašpar (ČVUT FIT) Data Storage II. MI-POA, 2011, Lecture 4 11/22 Tape Technologies Tape Terminology DAT roadmap DAT 160 DAT 8th Gen DAT 7th Gen DAT 72 DDS4 Longer Wide Tape DDS3 DDS1 DDS2 DDS2 Read/Write Compatibility Media Type Native Capacity Native Transfer Rate 90mTape 90m 120m Tape Tape 125m Tape 150m Tape 4 GB 12 GB ≤720KB/s ≤1.5MB/s 2 GB 183KB/s 1990 Pavel Tvrdík, Jiří Kašpar (ČVUT FIT) 1993 1996 170m Tape 150m Wide Tape Longer Wide Tape Longer Wide Tape 20 GB 36 GB 80 GB ~150 GB ~300 GB 1- 3 MB/s ≥3 MB/s ≥5 MB/s ≥8 MB/s ≥16 MB/s 1999 2003 2007 Data Storage II. Approx +2 years per generation MI-POA, 2011, Lecture 4 12/22 Tape Technologies Tape Terminology LTO Ultrium Drive Roadmap Generation 6 3.2TB Up to 270 MB/s WORM Capacity per cartridge and maximum limit for performance, without data compression Generation 1 100GB Up to 20 MB/s 2000 Generation 2 200GB Up to 40 MB/s 2002 Pavel Tvrdík, Jiří Kašpar (ČVUT FIT) Generation 3 400GB Up to 80 MB/s Generation 4 800GB Up to 120 MB/s Generation 5 1.6TB Up to 180 MB/s WORM WORM WORM 2004 Data Storage II. Approximately every two years MI-POA, 2011, Lecture 4 13/22 Tape Technologies Tape Libraries Tape Libraries Tape Library is a device consisting of: • one or more tape drives, • robotic hand, and • a magazine for storing tape media. A robot device is connected to hosts using the same interface technology as tape drives (SCSI, SAS, or FC). Robot moves tape cartridges among drives, slots in a magazine and I/O slot in accordance with SCSI commands passed to its LUN. A cartridge is identified by barcode and/or slot number. A robotic hand is equipped with a barcode reader. Pavel Tvrdík, Jiří Kašpar (ČVUT FIT) Data Storage II. MI-POA, 2011, Lecture 4 14/22 Tape Technologies Virtual Tape Libraries Virtual Tape Libraries Virtual Tape Library is an disk array with a tape library emulation firmware. It emulates all tape library functions and components using disk storage. Advantages: • no cabinet to drive delays, no load time, • very fast seek time, • more reliable than real tape library. Disadvantages: • no way how to move written data cartridge to to an archive for long time. Pavel Tvrdík, Jiří Kašpar (ČVUT FIT) Data Storage II. MI-POA, 2011, Lecture 4 15/22 Optical Storages Optical Storages • Optical Media Formats • Optical Libraries Pavel Tvrdík, Jiří Kašpar (ČVUT FIT) Data Storage II. MI-POA, 2011, Lecture 4 16/22 Optical Storages Preview of Optical Storage Media A myriad of Optical Technology: • • • • • • • • CD-DA (the basis of all other CD formats) CD-ROM CD-I DVI CD-XA MD (Mini Disks) CD-WO, CD-MO, WORM (Write Once Read Many) DVD Pavel Tvrdík, Jiří Kašpar (ČVUT FIT) Data Storage II. MI-POA, 2011, Lecture 4 17/22 Optical Storages Optical Libraries Optical Library is similar to a Tape Library, but uses optical drives and medias instead of tapes. As Optical device is random-access (not sequential) device, so data can be available fast (in seconds). Because of higher capacity and faster data access, optical libraries are being replaced by disk storages where possible. CD (not DVD) libraries are still good solution for long term archiving because of guaranted ~20 years of data readability. Pavel Tvrdík, Jiří Kašpar (ČVUT FIT) Data Storage II. MI-POA, 2011, Lecture 4 18/22 Flash Storage Flash Storage • Nonvolatile semiconductor storage o 100× – 1000× faster than disk o Smaller, lower power, more robust o But more $/GB (between disk and DRAM) Pavel Tvrdík, Jiří Kašpar (ČVUT FIT) Data Storage II. MI-POA, 2011, Lecture 4 19/22 Flash Storage Flash Types NOR flash: bit cell like a NOR gate • Random read/write access • Used for instruction memory in embedded systems NAND flash: bit cell like a NAND gate • Denser (bits/area), but block-at-a-time access • Cheaper per GB • Used for USB keys, media storage, … Flash bits wears out after 1000’s of accesses • Not suitable for direct RAM or disk replacement • Wear leveling: remap data to less used blocks Pavel Tvrdík, Jiří Kašpar (ČVUT FIT) Data Storage II. MI-POA, 2011, Lecture 4 20/22 Hierarchical Storage Systems Hierarchical Storage Systems Different storage technologies has different parameters: • price per stored Megabyte • access time To optimize price/performance ratio or storage behavior, more storage types are combined to single hierarchical system. The structure can be hidden in virtualized device, volume manager or file system. Pavel Tvrdík, Jiří Kašpar (ČVUT FIT) Data Storage II. MI-POA, 2011, Lecture 4 21/22 Sources Sources Wasfi G. Al-Khatib: SWE 423: Multimedia Systems. Chapter 8: Optical Storage Media. http://faculty.kfupm.edu.sa/ics/lahouari/Teaching/StNotes04.ppt Anxiao (Andrew) Jiang: Introduction to Flash Memories. CSCE 689: Frontiers in Storage Systems. Texas A&M University, August - December 2010. http://faculty.cs.tamu.edu/ajiang/689_01.ppt Henk Wubbolt: SAN Architecture BRU Product Choices. HP Advanced SAN Architecture, Prague, Czech Republic, 10..14 & 17..21 March - 2008 HP internal. Pavel Tvrdík, Jiří Kašpar (ČVUT FIT) Data Storage II. MI-POA, 2011, Lecture 4 22/22